Get extra lift from AOPA. Start your free membership trial today! Click here

Safety Pilot Landmark Accident: Unintended consequences

How Colgan 3407 changed the face of the industry

The aftermath of many accidents results in little change to the system, although a few alter the face of the industry.

Illustration by Brian Hubble

The aftermath of many accidents results in little change to the system, although a few alter the face of the industry. This particular tragedy is stunning for its simplicity in what happened—a basic airmanship failure. But a long chain of events led up to that individual failure and showed that fundamental systemic transformation was needed. It also serves as an example of how well-intentioned political pressure can lead to unintended consequences.

The flight
Colgan Airlines Flight 3407 departed from Newark Liberty International Airport (EWR), in New Jersey, about two hours late at 9:18 p.m. Eastern time on February 9, 2009, en route to Buffalo, New York (BUF). The aircraft that evening was a Bombardier DHC–8-400 (Q400), a late-model, high-wing twin turboprop. The flight proceeded normally until intercepting the localizer for the ILS Runway 23 approach into Buffalo.

The weather was typical for a Northeast winter evening with light snow and fog. There were airmets and some pireps for icing but nothing to preclude a routine airline flight.

At 10:04 p.m. the captain, who was the pilot flying, briefed the approach and the flight was cleared down to 6,000 feet. Passing 10,000 feet the sterile cockpit rule went into effect, which only allows discussion relative to flight operations. The crew violated that guidance when an appropriate conversation pertaining to ice on the airframe morphed into a discussion of the first officer’s overall icing experience and the time needed to upgrade to captain.

At 10:12:44 p.m. the autopilot’s altitude hold mode became active as the airplane approached the preselected altitude of 2,300 feet. Level-off occurred at 10:14:30 p.m. at 180 knots. At 10:15:06 p.m. the captain called for 5 degrees of flaps and shortly afterward ATC provided a heading of 260 degrees to intercept the ILS 23 localizer. The first officer acknowledged and the captain reduced engine power to just above flight idle at 10:16:02 p.m.

The final communication with Colgan 3407 occurred as ATC provided the handoff to the Buffalo tower. The cockpit voice recorder (CVR) noted sounds of landing gear movement, and the flight data recorder (FDR) showed the propeller condition levers were moved forward to maximum rpm while the autopilot began applying nose-up pitch trim.

Angle of attack (airspeed) was about to become critical. At 10:16:21 p.m., with the landing gear down, the speed decreased to 145 knots while the autopilot added additional nose-up pitch trim and an “ice detected” message appeared on the engine display. The captain called for 15 degrees of flaps and the before-landing checklist. The FDR showed the flaps moved to 10 degrees and airspeed dropped to 135 knots.

At 10:16:27.4 p.m., the CVR recorded the sound of the stick shaker. (This warns of an impending wing stall through vibrations on the control column and a rattling sound.) The autopilot disconnected at 131 knots and the disconnect horn sounded. At 10:16:27.8 p.m. (about one-half second later) the control column was pulled aft by the captain, and the engine power levers were advanced significantly one second later. FDR data measured engine power at about 75 percent torque, not quite full power.

The Q400 pitched up, rolled 45 degrees left, and then rolled to the right. Passing through wings level, the stick pusher activated at 10:16:34 p.m. (The stick pusher pushes the control column forward to decrease angle of attack.) The FO selected flaps 0 degrees at 10:16:37 p.m. and advised the captain. Airspeed was now down to 100 knots. The roll angle reached 105 degrees right wing down before the airplane began to roll back to the left and the stick pusher activated a second time at 10:16:40 p.m. The pitch angle was minus-1 degrees.

The aircraft went through several pronounced pitch and roll excursions during the next several seconds, with the FDR showing significant aft pressure on the elevator. The first officer asked about raising the landing gear, to which the captain replied, “Gear up” followed by an expletive. The stick pusher activated a third time at 10:16:50 p.m. with the captain stating, “We’re down,” and the sound of a thump. The airplane hit a house at 10:16:54 p.m.

Four crew members, 45 passengers, and one person in the house died in the crash. There were no survivors. Elapsed time from the first stick shaker activation to impact was about 23 seconds.

The aircraft
The 2008 Bombardier DHC–8 Q400, with a capacity of 74 passengers, had accumulated only 1,819 hours. It was equipped with a sophisticated flight management system (FMS) and glass instrument panel. The stall-protection package consisted of a stick shaker designed to alert pilots of an impending stall, and a stick pusher that would lower the aircraft’s angle of attack if the pilot failed to respond to the shaker. The Icing Ref switch programs the stick shaker/pusher mechanism to activate approximately 15 to 20 knots faster than normal to compensate for extra weight and wing contamination.

The VREF (reference landing speed) programmed into the FMS for the estimated landing weight of the accident aircraft was 118 knots. However, had the crew inserted the word “icing” into the system, 20 knots would have been added to the landing speed (at 15 degrees of flaps) for a VREF of 138 knots. So the aircraft anti-stall system was configured for icing, but the reference airspeed bugs were configured for an uncontaminated wing.

The airspeed indicator on the left of the primary flight display has a trend-vector tape to show where airspeed will be in 10 seconds and a low-speed warning cue, shown in red, that indicates when the stick shaker will activate.

No preimpact malfunctions were noted.

The terminal forecast at the time of the accident predicted winds from 250 degrees at 14 knots gusting to 24 knots, visibility 5 miles in light snow showers and mist, and an overcast ceiling at 2,000 feet.

There were numerous pilot reports for icing earlier in the day, with light-to-moderate and moderate rime icing from 3,000 to 14,000 feet. There were no urgent pireps, and pilots interviewed after the accident noted that this was normal mid-winter weather.

According to the NTSB, “The 9:54 p.m. METAR for Buffalo indicated winds from 240 degrees at 15 knots gusting to 22 knots, visibility 3 miles in light snow and mist, a few clouds at 1,100 feet agl, ceiling broken at 2,100 feet agl, overcast at 2,700 feet agl, temperature 1 degree Celcius, dew point minus 1 degree C, and altimeter 29.79 inches of mercury (Hg).”

The captain, age 47, held an airline transport pilot certificate and a first class medical certificate. He was type-rated on the DHC–8 in November 2008 after upgrading from the Saab 340. He had accumulated 3,379 hours of total flying time, including 3,051 hours in turbine airplanes with 1,030 hours as a pilot in command, but only 111 hours on the Q400.

The captain’s training records showed he received four certificate disapprovals including his instrument rating in 1991, commercial pilot in 2002, multiengine in 2004, and ATP in 2007. Only one failure was noted in his application to Colgan. There were several failures in proficiency checks while at Colgan. However, the check airmen who flew with the captain said his decision making was good and he was methodical. First officers who flew with him in February prior to the accident said he “handled the airplane well, used checklists, and did not miss call-outs” and “created a relaxed cockpit atmosphere but adhered to the sterile-cockpit rule.”

The first officer, age 24, held a commercial pilot certificate and a first class medical certificate. She was type rated (second-in-command privileges only) in the DHC–8 in March 2008. She had accumulated 2,244 hours of total flying time, including 774 hours in turbine airplanes, all on the Q400. She has worked for several months as a flight instructor for an FBO and for a year instructing at an airline training center.

She had one flight check failure for an initial CFI certificate, a not-unusual occurrence. She was described by check airmen and captains as “a good pilot who was sharp, assertive, and thorough; average to above-average for her level of experience; as a monitoring pilot was always ahead of the airplane and cross-checked her actions.”

The NTSB said, “Other captains indicated that, because of her abilities, the first officer could have upgraded to captain. None of the captains interviewed after the accident reported any problems with the first officer’s adherence to sterile cockpit procedures or stated that the first officer had made any unprompted configuration changes to the airplane while they were the flying pilot.”

The captain had commuted from Florida and slept in the crew lounge while the first officer commuted all night from her home on the West Coast to take a morning flight which was cancelled because of high winds. She had been up for hours and claimed to be suffering from a cold. It’s likely neither pilot was especially alert given the length of their duty day and the quality of rest that preceded it.

Probable cause
“The National Transportation Safety Board determines that the probable cause of this accident was the captain’s inappropriate response to the activation of the stick shaker, which led to an aerodynamic stall from which the airplane did not recover. Contributing to the accident were (1) the flight crew’s failure to monitor airspeed in relation to the rising position of the low-speed cue, (2) the flight crew’s failure to adhere to sterile cockpit procedures, (3) the captain’s failure to effectively manage the flight, and (4) Colgan Air’s inadequate procedures for airspeed selection and management during approaches in icing conditions.”

Clearly, the greatest part of this tragedy resides with the captain, whose airmanship and cockpit discipline left much to be desired. Monitoring airspeed during an approach is basic, as is stall recovery. The NTSB issued a highly detailed report of 187 pages, not including appendices, and a video re-creation which can be found online.

But there were significant systemic factors which the NTSB addressed with multiple recommendations. Here is a partial synopsis with emphasis on what may be most applicable to light GA pilots.

Automation can be a hindrance. When there is any confusion, usually it’s best to go down in levels of automation. Here, the autopilot maintained altitude until the airspeed decreased enough to trigger the stick shaker, preceding the stall. Navigation and altitude hold modes are not high levels of automation, but remember that the equipment does exactly what it is programmed to do. Changing configuration affects airspeed, but the captain, as pilot flying, missed that. Lowering the gear, adding flaps, and putting the condition levers (props) into flat pitch significantly increased drag. The first officer, as the pilot monitoring, did not catch it either.

This crew was complacent and distracted. These conditions are evident in many accidents. The sterile cockpit was developed for just this reason. In light aircraft, consider going sterile within 10 miles of the airport—inbound or outbound. Focus on the flying and traffic. As for complacency, aircraft are like chain saws; they cut both ways and neither behave well when mishandled.

To break a stall, altitude often must be sacrificed. At the time of the crash, many airlines trained pilots to make minor pitch changes and power out of stalls with minimum or no altitude loss. But power is destabilizing and if the stall is not fully broken, directional control problems and secondary stalls can rapidly develop. Break the stall first, and then come in smoothly with power. Had the captain done that we wouldn’t be having this discussion.

The captain’s training record was poor before he got hired by the airline, and not stellar afterward. One is occasionally allowed a bad day, but not consistently. In commercial aviation, the training and standardization programs are there to train, retrain, and ultimately remove those individuals who cannot consistently meet high standards. Colgan didn’t manage that well. For GA pilots, if you’re having difficulty flying well, take action! Invest in quality training to consistently meet at least private pilot standards. If that doesn’t work, maybe it’s time to give up the aeronautical chain saw.

The crew apparently did not understand the effects of the Ice Ref speed switch and apparently only Bombardier knew that the Q400 did not have a tail stall issue. NASA’s tail stall video, while not considered a factor by NTSB, needs to be understood in the context of the aircraft you are flying.

Fatigue was not mentioned as a factor, but then-NTSB Chairman Deborah Hersman noted that in her remarks. I agree. While the airlines allow their crews to commute, they also put the burden on the crewmembers to present ready for flight. Human nature being what it is, periodically somebody is going to show up exhausted. To deal with at least part of the fatigue problem, the commuting quandary should be addressed. According to the NTSB, more than two-thirds of Colgan pilots based in Newark commuted, either because of lifestyle or affordability. Not many accidents have been directly tied to fatigue—but it’s an overlooked link in too many, in my opinion.

The victims’ families somehow got the idea that the first officer was a major contributor to the accident and that low-time first officers were a significant risk. Tremendous pressure was put on Congress to force the FAA to raise the standards for first officers, with emphasis on total flight time as well as type of training. As a result, the FAA significantly increased the requirements—and cost—for one to become an airline first officer.

The first officer was not mentioned as a proximate cause, and in any case, she would have far exceeded the newly implemented requirements for both time and type of training. The AOPA Air Safety Institute reviewed airline accidents for 10 years preceding Colgan 3407 and found no involvement of first officers. In the hundreds of thousands of flight legs flown since Colgan, until the new rule went into effect this August, there were no accidents attributable to first officers.

One effect of the new rule was to decrease the supply of crewmembers, which led to service being cut at some smaller airports. It ultimately may have an effect on wages, which by any standard are absurdly low. That issue should be addressed on its merits, and not through the subterfuge of flight experience requirements.

Bruce Landsberg is the former president of the AOPA Foundation, which funds the work of the Air Safety Institute.

> Captain apparently does not understand aircraft systems and that the autopilot will pitch aircraft into an approach to stall when at greatly reduced power.
> Sterile cockpit procedures were not followed so crew is distracted.

Digital Extra: View an animation of this accident.

Related Articles